Expressive Threat Detection Validation Framework
Abstract
Under this project IET has developed a performance evaluation laboratory that consists of a synthetic dataset generator, a postprocessor to load generated data into relational databases, and a hypothesis scorer. The objective of this effort was to generate synthetic data sets for the counter terrorism domain to support objective performance of existing threat detection tools/technology. The generator can generate a file with 100,000 individuals and 1,000,000 observable transactions in about 12 minutes. The generator has over 100 different parameters that may be varied to make the link discovery problem easier or harder in specific ways. This massively parameterized problem space supports flexible experimentation that can address the following kinds of questions: What problem characteristics most influence a given technology's performance? What is the observed capabilities envelope of a given technology across the parameterized problem space? What performance can we predict for a given technology on a dataset with given characteristics?
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 01, 2006
- Accession Number
- ADA453605
Entities
People
- Brandon Goldfedder
- Robert Schrag